 |
ML Newsnew | past | best | rss | submit | about |
|
Stories from June 4, 2025 (UTC)
|
| 1. | Not all tokens are meant to be forgotten (arxiv.org) |
| 54 points by MarcoDewey on Jun 4, 2025 | hide | past | pdf | 23 comments | |
| 2. | Auto-Labeling Data for Object Detection (arxiv.org) |
| 6 points by nwlotz on Jun 4, 2025 | hide | past | pdf | 1 comment | |
| 3. | Esoteric Language Models (arxiv.org) |
| 3 points by frozenseven on Jun 4, 2025 | hide | past | pdf | discuss | |
| 4. | Generating High-Performance Tensor Operators with Hardware Primitives (arxiv.org) |
| 2 points by PaulHoule on Jun 4, 2025 | hide | past | pdf | discuss | |
| 5. | Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents [pdf] (arxiv.org) |
| 1 point by _tk_ on Jun 4, 2025 | hide | past | pdf | discuss | |
| 6. | Leancode: Understanding Models Better for Code Simplification of Pre-Trained LLM (arxiv.org) |
| 1 point by PaulHoule on Jun 4, 2025 | hide | past | pdf | discuss | |
| 7. | Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces (arxiv.org) |
| 1 point by felineflock on Jun 4, 2025 | hide | past | pdf | discuss | |